Re: forcedeth wierdness
Dean Hamstead wrote: thanks for that, it was hiding up in eth5 for some reason *shrug* and dmesg didnt bother to tell me i will have to set it to a more reasonable eth number Dean snip: forcedeth NIC missing Hi Dean This could be a udev persistent-net issue: When I upgraded my firewall box with three NICs from Sarge to Etch, one of them apparently changed hardware address and thus suddenly appeared as eth3 instead of eth0. Perhaps your /etc/udev/rules.d/z25_persistent-net.rules contains wrong or outdated entries too? Cheers, Jonas -- This message has been scanned for viruses and dangerous content by MailScanner, and is believed to be clean. -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Fwd: Re: Opinion question (Core2 Duo)
I have just completed unrestricted MP2/6-31G* energy procedure (after DFT/M05-2X) for a 98-atoms (first row) molecule in 19 hours with four-node amd64 Debian amd64 etch, NWChem suite. I understand (if I understand correctly) from your email that should I have had Core 2 I would not have had the time to take a coffee in between launching the MP2 procedure and getting the computational results. Interesting. Thanks francesco pietra --- Jo Shields [EMAIL PROTECTED] wrote: Subject: Re: Opinion question (Core2 Duo) From: Jo Shields [EMAIL PROTECTED] To: debian-amd64 debian-amd64@lists.debian.org Date: Tue, 18 Sep 2007 15:51:34 +0100 On Tue, 2007-09-18 at 10:41 -0400, Lennart Sorensen wrote: On Tue, Sep 18, 2007 at 03:08:48PM +0100, Jo Shields wrote: Or more? Buy an Altix! ;) Ehm, well the Altix uses either the itanium (why would anyone want that crap) or a dual socket core 2 based cpu. That hardly matches a 4 or more cpu opteron server. Let's assume I have large examples of both IA64 and AMD64. Plus further benchmark data we collected ourselves. IA64 is fast, for floating point code. On paper, it offers the same per-core-per-Hz FLOP count as Core (twice that of AMD64). And in practise, Altix scales, whilst the competition, well, doesn't. In our benchmarks, IA64 was not only faster per-GHz than POWER5 or AMD64, but faster in absolute terms too, with an 8-way test absolutely dominated by a 1.6-GHz-Montecito-based Altix, whilst AMD64 didn't even register a pulse. However, for IA64, compiler choice is key - using GCC to compile test code isn't just crippling the system, it's throwing away hundreds of thousands (if not more) of investment SGI has nothing of any real interest. No wonder they went under not that long ago. :) They've got SMP machines that don't choke at 4 cores. For some applications, that's of great interest. -- __ / Jo Shields [EMAIL PROTECTED] \ | Systems Manager, | \ Oxford Supercomputing Centre / --- \ ,__, \ (oo)___ (__))\ ||--|| * -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED] Check out the hottest 2008 models today at Yahoo! Autos. http://autos.yahoo.com/new_cars.html Don't let your dream ride pass you by. Make it a reality with Yahoo! Autos. http://autos.yahoo.com/index.html -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinion question (Core2 Duo)
On Tue, 2007-09-18 at 18:29 -0400, Douglas A. Tutty wrote: On Tue, Sep 18, 2007 at 11:16:14AM -0400, Lennart Sorensen wrote: Well if you need something to do floating point, then x86 isn't generally where you want to be. And yes if performance matters gcc is not what you want to use either. Is there a free alternative to GCC? Free? None worth using. Gratis? ICC, but only if you're unemployed. Look at it this way though. On a 64-node cluster, you'd be paying an extra £80 per machine (i.e. £5120) for a single chip-speed bump of about 16%. You'd pay an order of magnitude less for a decent compiler, which will get significantly better performance increases than the cpu speed bump. Nobody likes proprietary software (especially us sysadmins who need to make it work) but it's basic economics - 16% for £5k or 100% for £0.5k Where would you go for floating point? Last time I looked, Cray used Opterons as nodes in its supercomputers. Once you go beyond the desktop, a major factor becomes scalability, and ease of programming. Best bang-per-buck performance comes from a cluster of dual-dual-core Xeon nodes, with a reasonable message-passing interface like Infiniband. However, MPI programming is awkward (and in some problems sets not possible). In the land of SMP or Vector, you look at application scalability - if a 16-core Opteron system like a Tyan VX50 shows 0% improvement moving from 8 to 16 cores with a quantum chemistry code, and an Altix shows ~95% improvement, it's a no-brainer as to which is the better system to pick for that code. So for *small* codes, Xeon is a floating point monster. If you want scalable, you need to run a few benchmarks and decide what's an important factor for you (is your code parallel enough that a 8, 16, 32 or even 128 core limit per machine is a problem?). Then look at cost, choke on your coffee, and go back to buying Xeon clusters -- __ / Jo Shields [EMAIL PROTECTED] \ | Systems Manager, | \ Oxford Supercomputing Centre / --- \ ,__, \ (oo)___ (__))\ ||--|| *
Re: Fwd: Re: Opinion question (Core2 Duo)
On Wed, 2007-09-19 at 00:08 -0700, Francesco Pietra wrote: I have just completed unrestricted MP2/6-31G* energy procedure (after DFT/M05-2X) for a 98-atoms (first row) molecule in 19 hours with four-node amd64 Debian amd64 etch, NWChem suite. I understand (if I understand correctly) from your email that should I have had Core 2 I would not have had the time to take a coffee in between launching the MP2 procedure and getting the computational results. Interesting. Xeon/Core2 cheats slightly - it has combined units for both adding and multiplying on the chip - meaning if you do a multiplication immediately followed by an addition, then it'll happen in 1 cycle instead of 2, hence the twice as many FLOPs thing. Honestly, in desktop applications, it doesn't matter much - but in matrix operations, as used by most chemistry packages such as Gaussian or NWChem, it makes a full impact. So yes, you'd probably see a ~90% speed boost moving from an AMD64 to a Core2 of identical clock speed. If you have access to the NWChem source (you might not, I don't think we have any login credentials lying around to check with) you would see even bigger improvements with a commercial compiler and math library - you may find a cheaper option to improve performance than buying a new Core2 rig is to buy Portland C (or Pathscale C), and link against the free AMD Core Math Library instead of conventional open-source BLAS/LAPACK routines. -- __ / Jo Shields [EMAIL PROTECTED] \ | Systems Manager, | \ Oxford Supercomputing Centre / --- \ ,__, \ (oo)___ (__))\ ||--|| * -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Testing a hardware
Hi, we have a new box here that seems a nice machine. It has: - gigabyte ga-m61p-s3 mobo with: nVidia Corporation MCP61 Memory Controller (rev a1) Audio device: nVidia Corporation MCP61 High Definition Audio (rev a2) nVidia Corporation MCP61 SATA Controller (rev a2) nVidia Corporation Unknown device 03d0 (rev a2) FireWire (IEEE 1394): Texas Instruments TSB43AB23 IEEE-1394a-2000 Controller (PHY/Link) It has a sata hd and 2 memory panels (1gb) of kingstom (800MHz) . The kernel 2.6.18 from debian etch recognizes all the hw and install boots without any problem. Also, the 2.6.22 from sid. Howeber, I have one big problem with this box: crash many times . I can boot, look the login screen, do nothing, and then the image get distorsioned and the box crash. I can boot the box, make a ssh, make ls and the box crash. I can boot the box, make an ssh, compile a kernel, a lot of things without any problem, make a logout, and the next day the box is hang. I have not any messages in the /var/log/kern or syslog Also, I have some annoying thing. In theory the mobo supports the memory at 800Mhz. The memory that I have, also support the 800Mhz. But, although I configure the voltages as the manual of the mobo and the memory says, I got never have a message from the bios saying that the memory is 800. Always says 667 So, my question is? How I can determinate if: - the memory fails (memtest done, but only 2 hours ..) - mainboard fails - debian kernel stock fails? - my fingers fails and I have a misconfigured bios ... Regards, Leo -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Testing a hardware
A Dimecres 19 Setembre 2007 15:50, Daniel Tryba va escriure: On Wed, Sep 19, 2007 at 01:30:43PM +0200, Leopold Palomo-Avellaneda wrote: [crashes] So, my question is? How I can determinate if: - the memory fails (memtest done, but only 2 hours ..) - mainboard fails - debian kernel stock fails? - my fingers fails and I have a misconfigured bios ... Did you try booting with various apic/acpi settings (noapic/noacpi/nolapic/etc.)? nops, but I have to try. But this could be so important? Leo -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Testing a hardware
On Wed, Sep 19, 2007 at 01:30:43PM +0200, Leopold Palomo-Avellaneda wrote: [crashes] So, my question is? How I can determinate if: - the memory fails (memtest done, but only 2 hours ..) - mainboard fails - debian kernel stock fails? - my fingers fails and I have a misconfigured bios ... Did you try booting with various apic/acpi settings (noapic/noacpi/nolapic/etc.)? -- Daniel Tryba -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinion question (Core2 Duo)
Douglas A. Tutty píše v St 19. 09. 2007 v 00:31 +0200: Why is it that debian doesn't do pre-emption in the kernel? I can imagine workloads (typically on servers) where preemptive kernel is not necessary (or even can be bad for performance). Vit -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]
Re: Opinion question (Core2 Duo)
Douglas A. Tutty píe v St 19. 09. 2007 v 00:31 +0200: Why is it that debian doesn't do pre-emption in the kernel? I can imagine workloads (typically on servers) where preemptive kernel is not necessary (or even can be bad for performance). Vit Here's some info on pre-emption and the kernel, may be of interest.. http://www.linux-foundation.org/en/Linux_Weather_Forecast#Real-time_preemption Norv Sick of deleting your inbox? Yahoo!7 Mail has free unlimited storage. http://au.docs.yahoo.com/mail/unlimitedstorage.html
Re: Testing a hardware
Leopold Palomo-Avellaneda wrote: A Dimecres 19 Setembre 2007 15:50, Daniel Tryba va escriure: On Wed, Sep 19, 2007 at 01:30:43PM +0200, Leopold Palomo-Avellaneda wrote: [crashes] So, my question is? How I can determinate if: - my fingers fails and I have a misconfigured bios ... You could try adjusting your RAM timings. Your RAM manufacturer may have stated a default setting for timings in its specs. Bill -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of unsubscribe. Trouble? Contact [EMAIL PROTECTED]